1. Field of the Invention
The present invention relates to a technique of automatically capturing an image by recognizing an object status from an image.
2. Description of the Related Art
There has conventionally been known a technique of automatically capturing an image by recognizing an object status from an image. As a concrete technique, an image is automatically captured when the object status is recognized from an image to detect that the object faces the front or takes a predetermined action, posture, pose, or facial expression (see Japanese Patent Laid-Open No. 2001-51338).
As another technique, an image is shot when an object stands still for a predetermined time. Alternatively, when an object stands still for a predetermined time, he is given a sign to start shooting. If the object keeps still for another predetermined time, an image is shot (see Japanese Patent Laid-Open No. 2003-107574).
To recognize the state of a specific object such as a person after detecting the specific object, more detailed information is extracted from image data of the specific object. In this case, however, recognition processing may not be executed accurately under the influence of various factors.
For example, recognition fails when the face size of an object is small and there is little information for recognizing a face expression and state, or when recognition is affected by the shadow of illumination or accessories such as sunglasses. When a plurality of objects are to be photographed, the same recognition processing is executed a plurality of number of times. Recognition processing sometimes fails for a given object, and no automatic image capturing is executed.
If no image recognition is accurately achieved under the influence of a variety of factors, no image may be shot though automatic image capturing conditions are satisfied in practice. In some cases, a very long time is taken to perform automatic image capturing.